DOI 10.1393/ncc/i2012-11220-y Colloquia: TOP2011
Searches for resonant production of top anti-top quarks
at the Tevatron
V. Boisvert(∗) for the CDF and D0 Collaborations Royal Holloway, University of London - Egham, Surrey, UK ricevuto l’ 1 Marzo 2012
pubblicato online il 28 Maggio 2012
Summary. — A review of the measurements pertaining to searches for resonant
production of top and anti-top quarks from the CDF and D0 Collaborations is presented, using different methods to reconstruct the top anti-top quark invariant mass.
PACS 14.65.Ha – Top quarks. PACS 12.60.Nz – Technicolor models.
1. – Introduction and motivation
As the heaviest particle known so far, the top quark participates in many beyond the standard model (SM) scenarios. Such scenarios might include extended gauge theories with massive Z-like bosons and Kaluza-Klein (KK) states of gluons, weak bosons and gravitons. Some of these models predict an enhanced coupling to third generation quarks and t¯t is the dominant decay mode. A resonant production of a top and anti-top quark
would appear as an unexpected structure in the spectrum of the invariant mass of t¯t
pairs Mt¯t.
Models are grouped according to the spin of the object, its color content and CP parity [1]. There can also be non-trivial interferences effects between new resonances and SM t¯t production. Experimentally the approach is to obtain the reconstructed Mt¯t
and to look for any deviation from the SM t¯t production. Most results extract limits for
a topcolor leptophobic Z of narrow width (spin 1 and color singlet), while one result quotes a limit for a massive gluon production (spin 1, color octet).
Looking at the current differential cross-section of SM t¯t production [2], there are very
few events found at very large Mt¯t values, making the signal of a resonant production
at high mass potentially possible to observe. However, current CDF and D0 results are focused on a non-boosted topology for the final state, which is less efficient at very large
(∗) Formerly from University of Rochester - Rochester, NY, USA. c
2. – Lepton+jets event selection
As four out of the five reported results make use of the lepton+jets channel (where one of the W decayed leptonically, while the other decayed into hadronically), the event selection is presented here. First, for the trigger selection, CDF requires an electron (muon) trigger with ET(pT) > 18 GeV while D0’s trigger requests an electron (muon)
and a jet. CDF requires an isolated electron (muon) with ET(pT) > 20 GeV, while D0’s
muon requirement is that the pT > 25 GeV and |η| < 2.0. The Missing ET (ET) is the
usual vector sum of calorimeter energy deposits, corrected for identified objects. The CDF requirement is that for both electron and muon ET > 20 GeV, while for D0 the
electron (muon) requirement isET > 20 (25) GeV. CDF uses the anti-kT with R = 0.4
algorithm for jet clustering while D0 uses the cone algorithm with R = 0.5. The jet pT
and eta requirements are analyses dependent. The identification of b-jets is done using the displaced vertex algorithm SECVTX for CDF which has an identification efficiency of 50% for b-jet and 2% for light jets. D0 uses its neural network (NN) (discriminating variables are d0, mvtx, displacement significance and the number of tracks within the
secondary vertex) with a requirement of NN > 0.65, which results in a b-jet identification efficiency of 55% and a light jet efficiency of less than 1%. Finally, both collaborations apply some standard event cleaning requirements.
The backgrounds in this channel are the same that are involved in the t¯t cross-section
measurements. The most important ones are the W +jets production and the multijets production. Both are estimated using data-driven methods. The electroweak back-grounds (diboson, single top quark) are estimated using Monte Carlo predictions. 3. – CDF: Search for new color-octet vector particle
This analysis [3] uses 1.9 fb−1of Tevatron data and searches for a massive generic gluon
G. It is assumed that the coupling between the massive and massless gluon is 0, while
there can be interference between G and the process qq→ g → t¯t. The coupling of G to quarks is assumed to be parity-conserving. In t¯t production there are three observable
parameters: the product of massive gluon coupling strength λ = λqλt, the mass M and
the width Γ. The SM process gg → g → t¯t is a background. The Mt¯t is reconstructed
from the parton level momenta event by event using the Dynamical Likelihood Method, which was also used for a top mass measurement. A path integral is built, which includes a transfer function treated as a probability density function (pdf) from the observed to the parton kinematics, and each event is the average over all the possible paths. It is assumed that the transfer functions are independent of the t¯t production matrix and so
Fig. 1. – Upper and lower limits on the strength of the coupling with 95% CL as a function of the mass of the massive gluon. The limits become worse at the larger width.
into an unbinned maximum likelihood. The pdf contains a term which is the ratio of
g + G to SM t¯t production, which implies that Parton Density Functions (PDF), top
propagators, decay ME and final state densities will cancel out. Another term of the pdf is the resolution function which translates from reconstructed to true Mt¯t. Examples
of the excluded regions for the coupling parameter is shown in fig. 1. In summary, no significant indication of the existence of a massive gluon with|λ| > 0.5 is observed in the search region of 400 GeV/c2< M < 800 GeV/c2and 0.05 < Γ/M < 0.5.
4. – D0: Search for top anti-top quark resonances using 3.6 fb−1
This analysis [4] uses the lepton+jets channel and obtains results separately for the 3-jet signal region and for the greater or equal to 4-jet region. The product of acceptance, efficiency and branching ratio after the event selection is 3.4% and 4.4%, respectively, for the SM t¯t events and 2.8% and 3.9% for events from a Z with mass of 650 GeV/c2. The
analysis obtains Mt¯tfrom using up to four leading jets and uses the W mass constraint
for obtaining the z component of the neutrino momentum. If two solutions are found, the smallest one is used and if no real solutions are found pzis set to 0. This procedure
provides better sensitivity at large values of Mt¯t than using a kinematic fit. Limits are
obtained using a Bayesian approach with a Poisson probability for the number of events in each bin, and a flat prior for σB. The Mt¯tdistribution in the greater than or equal to
four jet signal region is shown in fig. 2 alongside the exclusion limit plot. In summary, a topcolor leptophobic Z is excluded at 95% CL below 820 GeV/c2.
5. – CDF: Search for resonant production of t¯t pairs in 4.8 fb−1
This analysis [5] uses the lepton+jets channel and for each event the t¯t hypothesis
is applied by mapping the observed event kinematics to the parton level using the Ma-trix Element for t¯t production and decay. A pdf is constructed representing Mt¯t which
includes a sum over jet-parton assignments. The pdf also contains a transfer function mapping jets to partons. These functions are obtained from Monte Carlo and have 10 bins in jet ET and five bins in jet η. The probability for an event is a function of the
signal fraction and the signal and background templates. The likelihood is written as the product of the per-event probabilities. The total probability density is shown in fig. 3
the data points show the measured σX· B(X → t¯t) for a narrow resonance obtained for the full
data set. The shaded areas indicate the 68% and 95% fluctuation bands expected under the assumption of pure SM t¯t production.
alongside the expected and observed limit. In summary, a topcolor leptophobic Z is excluded at 95% CL below 900 GeV/c2.
6. – CDF: Search for resonant production of t¯t decaying to jets
This analysis [6] uses the all hadronic channel and 2.8 fb−1. The advantages of the all hadronic channel are the high branching ratio, the good mass resolution and the complementarity of this result compared to the lepton+jets channel. The all hadronic event selection starts with requesting a jet trigger (at least 4 jets of ET > 10 GeV, which
has a signal efficiency of 80%). It vetoes leptons andET and asks for either six or seven
jets with ET > 15 GeV and |η| < 2.0. The b-jet identification algorithm is SECVTX
described above. To suppress the overwhelming multijets background a neural net is used containing 10 variables that discriminate between multijet events and t¯t events.
A requirement of NN > 0.93 is applied. The leftover multijet background seeping into the signal region is estimated from the tag rate matrix from four or five jet bins and tested on various control samples, built from looser requirement on the NN value. A likelihood is calculated by integrating the signal matrix elements, which allows not only to calculate Mt¯tbut also to suppress the background. The Mt¯tdistribution is shown in
fig. 4 alongside the limit obtained. In summary a topcolor leptophobic Z is excluded at 95% CL below 805 GeV/c2.
Fig. 3. – Left: total probability density for the 1366 t¯t events observed in 4.8 fb−1. Right: expect-ed and observexpect-ed 95% CL upper limit for σ(p¯p→ Z)× BR(Z→ t¯t).
Fig. 4. – Left: Reconstructed Mt¯t distribution vs. the SM expectation. Right: Expected and
observed upper limits on leptophobic topcolor Z in 2.8 fb−1.
7. – CDF: A search for boosted top quark using 5.95 fb−1
This analysis [7] is the first observation of boosted top quarks at the Tevatron. The observation of massive collimated jets is an important test of perturbative QCD, helps tune the Monte Carlo event generators, gives insight into the parton showering mecha-nism, and opens the door to resonant top anti-top searches in the boosted topology. It is the first measurement probing the t¯t sample with top quark pT > 400 GeV/c. It is
interesting to note that for a Lorentz boost greater than three the top decay products collimate into a single massive jet. The jet trigger requested expects at least one jet with
ET > 100 GeV. The jets are reconstructed using the Midpoint algorithm with R = 1.0
and the calorimeter towers are turned into four-vectors using the so-called “E-scheme”. A jet energy scale correction as well as a multiple interaction correction are applied. At least one of such reconstructed jets with pT > 400 GeV/c and |η| < 0.7 is requested.
Fig. 5. – Left: the mjet2 vs. mjet1 distribution for all data events. Right: the SMET vs. mjet1
there is a slight correlation between those two variables, the expression used to obtain ND is Rmass= NBNC NAND , (2)
where Rmass = 0.89±0.03(stat.)±0.03(syst.). The estimated amount of multijet
back-ground is 14.6 ± 2.76. In the lepton+jets channel the ET significance required is
4 < SMET < 10. In this channel the variable mjet2 is no longer discriminating between
the multijet events and the t¯t events and so the four regions are defined in the SMET vs. mjet1 plane to extract the multijet background, see fig. 5. This estimate is 31.3± 8.1.
Combining the all hadronic and lepton+jets channel 57 data events are selected com-pared to an expectation of 46± 8.5(stat.) ± 13.8(syst.). An upper limit on the cross section is obtained by using a Bayesian treatment with a flat prior. The expected limit is
σ < 33 fb at 95% CL and the observed limit is σ < 38 fb at 95% CL. These numbers are
an order of magnitude larger than the SM predicted cross section, and are dominated by the background prediction. The analysis also makes use of the all hadronic channel to put a limit on the resonant top anti-top quark production: a limit on the cross-section,
σ < 20 fb at 95% CL is obtained.
8. – Conclusion
Several searches for resonant production of a top and anti-top quark at the Teva-tron were presented. The best limit obtained for a leptophobic topcolor Z is MZ >
900 GeV/c2 (for Γ = 1.2%). A search for a massive gluon was also presented and limits
put on its coupling strength. This massive gluon is generic although assumes parity conservation. The first boosted t¯t analysis at the Tevatron was also presented, providing
the most stringent limit on boosted top quark cross-section. REFERENCES
[1] Frederix R. and Maltoni F., JHEP, 0901 (2009) 047. [2] Jung A., these proceedings.
[3] Aaltonen T. et al. (CDF Collaboration), Phys. Lett. B, 691 (2010) 183. [4] Abazov V. M. et al. (D0 Collaboration), D0 Note 5882-CONF (2009). [5] Aaltonen T. et al. (CDF Collaboration), Phys. Rev. D, 84 (2011) 072004. [6] Aaltonen T. et al. (CDF Collaboration), Phys. Rev. D, 84 (2011) 072003. [7] Aaltonen T. et al. (CDF Collaboration), CDF Conf. Note 10234 (2011). [8] Kidonakis N. and Vogt R., Phys. Rev. D, 68 (2003) 114014.